As hyper-converged infrastructure makes its way from edge applications to core workloads, the scale of the implementations is increasing. With scale comes added pressure on the resources being consumed by the HCI system such as rack space, power and cooling, as well as the network infrastructure to which it is connected.
Depending on the age and current capacity of your existing network, large scale HCI installations may require a network upgrade to ensure there is no loss of performance, according to Matthew Oostveen, Chief Technology Officer, Asia Pacific, Dell EMC.
In an interview with Networks Asia, Oostveen talks about the importance of the network when deploying hyper-converged infrastructure, its uses cases, and issues that need to be addressed.
“For customers that need to overcome scaling issues, Dell EMC has engineered a Rack-Scale Hyper-converged system called the VxRack with in-build network topology. This gives the ability to scale HCI clusters without the issues associated with the appliance form factor that relies on a third-party network.
“At Dell Technologies, we are very excited about the promise that HCI holds for our customers in solving the IT challenges they face in the future. And whilst we are number one in the HCI market, we also see a place for the SAN. There are some very good use cases where SAN-based technologies will continue to provide the right solution for customers. This will ensure that SANs will have a place in the modern IT datacenter.”
Does there need to be any changes to IP address provisioning? How does hyper-converged lend itself to VMs and VM sprawl? Does scalability become easier with increased virtualization? What is the best way to address scalability? How are old technologies like load balancing and fail-over handled in hyper-converged infrastructure? Should we look towards the application workload itself to prioritize the traffic?
VM sprawl is an issue that has challenged the IT department since virtualization was popularized in the last decade. As we made it easier to provision VMs, we saw rampant VM sprawl as policy and process scrambled to catchup with the technology. Irrespective of the delivery platform – whether it is on premises or public cloud – discipline is required by the IT department to ensure sprawl is managed.
Managing sprawl is first and foremost a policy and process matter – as custodians of technology, the IT department must ensure they strike a balance with their users to enable frictionless service delivery while simultaneously managing appropriate usage.
On a technology level, the problem of VM sprawl is lowered through the use of HCI when compared to traditional infrastructure. Through the usage of policy based management and increased visibility of compute, networking and storage, sprawl can be reduced.
In the case of the VxRail, scalability is easy. Starting with an appliance of 3 nodes, a VxRail cluster can be upgraded in single node increments. The VxRail Manager bundled with the system allows the operator to rapidly build the new node – taking less than 5 minutes to go from powering on to completion of the build.
VxRail hardware is designed to achieve more than ‘5 nines’ of availability. Coupled with the features in vSAN and vSphere, it is possible to achieve more than ‘6 nines’ of availability for business-critical applications.
VxRail provides high availability and resiliency functionalities via vSphere features such as vMotion, DRS and HA (High Availability) for avoiding planned and unplanned downtime and site maintenance of a virtual environment. In addition, vSAN features Failure to Tolerate (FTT) and Fault Domains (FD) provides site level protection against disk, host, connectivity, power, and rack failure. For multi-site DR, vSAN stretched cluster provides active/active protection against zero data loss and near zero downtime between two sites. Additionally, through the VxRail app-store, Dell EMC Data Protection Suite for VMware can be added to VxRail.
Beyond things like VDI, what are other use cases? Are there instances when an enterprise should look at converged rather than hyper-converged infrastructure? Are things like DR and BCP changed?
The use case with the most interest from customers today is Hybrid Cloud on HCI. A few months ago, Dell EMC launched a new product that launched in this space. With VxRail EHC, the VxRail can now be ordered as a turn-key hybrid cloud platform. This product allows customers to acquire a fully built cloud platform with factory baked software and services that are tuned to the task being performed on the customer site.
HCI is incredible technology, but that doesn’t mean it is perfect for every use case we see. Reflecting this is data that shows the converged infrastructure market – or CI – is still much larger than HCI.
For customers that have very large use cases and are operating at higher economies of scale, CI such as VxBlocks are often the better choice. These systems are mostly deployed to run mission critical and business critical workloads of chief importance to the organization.
Examples of industries where we see strong use of CI includes banking and financial services, telcos, as well as large government institutions.
As hyper-convergence moves to eliminate storage, compute and networking silos, what reskilling is needed for IT staff? Do you have any regional examples of hyper-convergence deployed for the right reasons and correctly?
My advice to technical professionals considering HCI or CI is that the future is bright. Reskilling is fundamental to the role of an IT professional, but in the case of CI and HCI, there is a bigger opportunity to take on a more important role in the organization thereby future your career. On a personal level, I have followed the careers of several young engineers and witnessed their responsibility change when Vblocks or VxRails were added to their environments.
Systems engineers have evolved into DevOps Engineers where responsibilities evolved from routine systems maintenance to design and building of PaaS environments. In the case of DBAs, roles move into that of a data engineer and responsibilities move from upgrading and patching databases into insight visualization and “data wrangling”.
When it comes to deployment, the configuration flexibility of hyper-converged infrastructure plays a crucial role, as it enables us to customize the deployment to better meet business requirements.
In the case of our customer, KPIT Technologies (KPIT), a global IT consulting and product engineering organization based in India, we deployed our VxRail hyper-converged infrastructure appliance for KPIT’s development and test environments, and backup and recovery. With benefits such as an improved return on investment and flexibility to integrate data protection across the organization, the VxRail solution not only enabled KPIT to grow new business incrementally, but also helped them gain business agility and flexibility to embrace digital transformation. KPIT achieved 35 percent additional cost savings, compared to a traditional or converged approach, in a short span of time.